13.1 Nordic leukaemia study revisited One of the most influential studies of the health effects of exposure to internal man-made radioactive isotopes was published in 1992, at about the time the discussions were taking place about how to fight the Sellafield leukaemia trial, Reay and Hope vs. BNFL. The prosecution lawyers were to have to explain how the 10-fold child leukaemia excess in Seascale near Sellafield could be caused by radiation from the plant. One explanation (the correct one, I believe) was that the ICRP risk model was incorrect for internal radionuclides. The other explanation was the one that the lawyers were advised to pursue, that it was caused by genetic damage following parental pre-conception radiation. This was bad advice, as it happens, and the case was lost. However, for the nuclear establishment this case was a potential disaster. They were well aware of the scientific weakness of the ICRP model for internal radiation. One of their experts told Bill Pritchard this over a pint during the Hinkley enquiry. Something was clearly necessary to block off this possibility and so something was done. Enter Dr Sarah Darby and Professor Richard Doll and the Nordic Leukemia study. I had already criticized this study in Wings of Death but because it reappeared as a major piece of evidence that internal radionuclides were safe at low doses during the deliberations of a new government Committee Examining Radiation Risk from Internal Emitters (CERRIE) in 20022003 I was to learn a great deal more about what went into the study, and what was left out. The story is quite entertaining, and gives an insight into the process of unravelling a system of scientific bias and the effort required to discover the truth. First let me outline what I knew at the time of writing Wings of Death. The title of the paper, which was published in the British Medical Journal in April 1992 , about a year before the trial began in earnest was Trends in childhood leukaemia in the Nordic Countries in Relation to fallout from nuclear weapons testing [Darby et al, 1992]. The authors were Sarah Darby, Jorgen Olsen, Richard Doll, Bharat Thakrar, Peter de Nully Brown, Hans H Storm, Lotti Barlow, Froydis Langmark, Lyly Teppo and Hrafn Tulinius. Most of these people were on the paper because they supplied the data. The calculations were done in Oxford. Results of a rather complicated time-series Poisson regression analysis involving a lot of mathematical manipulation and splicing together of data from Denmark, Sweden, Iceland, Norway and Finland indicated that there was little evidence of increased incidence in children born in the high fallout period. For children after birth leukemia incidence and red marrow dose were not related overall but rates of leukemia in the high exposure period were slightly higher than in the surrounding medium exposure period. Looking at the graphs of leukaemia rates standardised for age, sex and country it is difficult to see any change over the period of the weapons fallout. Fig 13.1.1 used the data from the paper to show the rates for the 0-4s. But close reading of the text shows that this graph is a construction, a creation of mathematics and hope. It does not actually show the rates in the 'Nordic Countries' because up to 1958, only rates from Denmark were available. This is the major and lethal flaw in the study. What Darby et al did was to sell this as a study of child leukaemia in a homogenous population over a span of 36 years, children who were exposed in the middle of the period to internal manmade radioactive materials from weapons fallout. In reality it was a study of at least two distinct populations. Before the fallout it was a time series of Denmark. After the fallout it was a time series of Denmark plus all the other Nordic countries, adjusted in various unelaborated and complicated ways. Anything that might have happened in Denmark after the fallout would be swamped by the big populations of the other Nordic countries. Anything that might have happened in the other Nordic countries would be missed because there was no pre-fallout baseline before 1960 to compare the rates with. The doses in the different countries are quite different due to patterns of rainfall and so the development of the illness would have a different lag time. The same process, carried out by IARC at Lyon has successfully smudged out increases in childhood leukemia after Chernobyl, another story.
Fig 13.1.1 The time series of the Nordic leukaemia study for 0-4s as reproduced in Wings of Death Also shown is the series for the Nordic countries with the Danish data removed. Note the apparent increase in leukaemia rates after 1959 if the early (Danish) data are cut out. In 1995, I wrote to Sarah Darby to ask if I could have a look at the raw data. I wanted to see what had happened to the Danish series after it had been absorbed into the general Nordic population. She replied on 1st September that the data was not available. I think it must have been destroyed when Dr Bharat Thakrar left our Unit. But my daughter Celi, who works as an epidemiologist at the London School of Hygiene and Tropical Medicine met Thakrar and spoke with him about the data. He told her that he certainly had not destroyed it. And there the matter would have ended had it not been for the CERRIE committee. I will give an account of the Committee Examining Radiation Risk from Internal Emitters in part IV as it is a major triumph of our campaign, but here I will now relate how it came about that the plot of the Nordic Leukemia study was finally unravelled. 13.2 Hakkulinen sees, fallout brings disease Since the remit of the CERRIE committee was to look at internal radionuclides, it was inevitable that someone would advance the Nordic leukaemia study as proof that there was nothing to worry about, and that is exactly what happened. In this case, interestingly and unhappily for her, Sarah Darby had been brought into the committee at a late stage (after its members had been chosen) as a 'neutral epidemiologist' by the Chair, Dudley Gooodhead. CERRIE itself had been forced into existence by the incontrovertible evidence of increases in infant leukaemia in the children who were in the womb at the time of the Chernobyl black rain. The effect occurred in five different countries and had to be an effect of Chernobyl; but if it were, then the ICRP risk factors were in error by about 100 to 500 times (which is about right, actually). The nuclear apologists on CERRIE fell back on the Nordic Leukemia study. Why was there no effect there, they cried. The study looked at child leukemia 0-4 but infant leukaemia would have shown up if there were an effect. So I countered with my arguments from Wings of Death. The series was bogus. The study population was changed in the middle of the series, at exactly the point something should happen. How was the series constructed, I asked Darby. Can we see the data? Can we look at Denmark only and see whether there was an effect there, in the only dataset that covers the whole period? No you can't, she replied. The data has been destroyed, removed, wiped long ago. It does not exist any more. But some data did exist, and some studies of Leukemia in Denmark had been made it seemed. I began to dig. And (strangely) thanks to Richard Wakeford of BNFL, we had our first break. Wakeford agreed to write to Jorgen Olsen of the Danish Cancer registry and ask for the raw data. But a few months after this he also turned up a fat report entitled Trends in cancer incidence in the Nordic Countries (Hakkulinen et al 1986). This gave data on adult, infant and childhood acute leukaemia aggregated in five year blocks. The authors noted that the only data that covered the period before 1960 was from Denmark They showed that for
acute leukaemia, there were significant increases in the incidence of infant, childhood and also adult leukaemia in Denmark over the period of the fallout. They noted (p92) Hansen et al. (1983) concluded that the increases in acute leukaemia in age groups 50 years and over may not be an artefact. They suggest environmental factors are involved. Ionizing radiation is a well known risk factor in leukaemia. They did not refer to the amazing increases in childhood acute lymphatic leukaemia plotted in their publication. I have plotted the rates of childhood acute leukaemia in Denmark given by the Hakkulinen et al publication in Fig 13.2.1. I became quite excited by this and obtained the paper by Hansen et al referred to in the Hakkulinen et al report. Sure enough, Hansen (1983) had examined leukaemia in Denmark and published in the prestigious Journal of the National Cancer Institute a paper entitled Trends in the incidence of leukaemia in Denmark 1943-47: an Epidemiolgic study of 14,000 patients. (Hansen et al, 1983) In the abstract, Hansen et al stated; An epidemiologic study of the total population of patients with leukaemia in Denmark during 1943 to 1977 was performed. The material stemmed from the National Danish cancer Registry and was believed to be complete. . . Over the 35 year period the incidence of acute leukaemia increased threefold in the age groups 0-9 and 5070+ years; whereas the increase in the age groups 0-9 climaxed in the period 68-72, the increase in the age group 50+ was sustained. Of course, a peak in incidence in 0-9 year olds in 1969 to 1972 points to an exposure in the womb or early childhood in 1960 to 1964, the fallout peak! These papers suggested that the Danish data showed an effect for acute leukaemia. Why had they not been referenced in the Nordic leukaemia study? After all, if you write a scientific or medical paper, you should include the work of other people in the area. I was confused and I decided to put some pressure on Darby and Doll to find their lost data. I contacted the Editor of the British Medical Journal and asked him to look into the matter of the Danish data, since the paper was very influential and had been published by his journal. I also wrote to the Danish Committee for the Investigation of Scientific Dishonesty in Copenhagen. I wrote to the Fitness to practice Directorate of the British Medical Association. The Hakkulinen and Hansen data suggested that the fallout had caused significant increases in acute leukaemia in Denmark in children and also in adults. Neither of these papers had been cited in the Nordic study which showed no increases in leukaemia in Denmark. What was going on?
Fig 13.2.1 Childhood acute leukaemia 0-4 in Denmark 1943-77 from Hakkulinen et al (1983) bars. Also shown is the Nordic leukaemia study trend (rates per 100,000) 13.3 The data magically appears
The Danish Committee on Scientific Dishonesty started digging. The British Medical Journal wrote to Darby and Doll. A miracle occurred. Dr Darby found a disc with the original data. This was sent to Prof Roger Robinson of the BMJ and passed on to me by the editor, who declared that he was satisfied that the explanation given by Darby and Doll for the discrepancies between their graph and that of Hakkulinen et al and Hansen. Let me remind you what is happening here. I am questioning the validity, direction and honesty of one of the most important studies ever published on a question at the basis of the operation of the nuclear project. The authors are the most respected people in epidemiology. Their co-authors are directors of the various Scandianavian cancer registries. Olsen is Director of the Danish Cancer Registry which was taken over by the State in 1997. He is presently the Head of the Institute of Cancer Epidemiology in Copenhagen with 40 researchers, 20 technical staff and a secretariat. Their research is based on extramural funding from the US National Cancer Institute and the International Agency for Research on Cancer in France and the Nordic Cancer Union (which he is also president of). Are they looking at the effects of radiation? Don’t bet on it. Do we think the Danish Committee for Scientific Dishonesty will take on these people? Forget it. The Nordic study is regularly used as a cornerstone of the nuclear industry case. In 1993 NRPB wrote to Bramhall using this study to deny the existence of radiation related bone cancer in Wales. There is now a massive cancer epidemic. There is a lot at stake. The Danish Committee judge, Henrik Waaben who had taken the case over wrote to me and said that he couldn’t investigate the paper because it was too long ago. A 5-year Statute of Limitations applied, apparently. I don’t blame him for chickening out. The explanation given by Darby and Doll was that Hakkulinen and Hansen looked at Acute Leukemia. Darby Doll et al looked at all All Leukemia and that in the early years before 1960, many children registered with leukaemia of unclassified type. The questions being asked implicitly and explicitly by Hansen and Hakkulinen in 1983 and 1986 had apparently already been addressed in the International Journal of Epidemiology in 1989. The paper Incidence of Childhood Cancer in Denmark 1943-84 was by Peter de Nully Brown, Henrik Hertz, Jorgen Olsen, Minna Yssing, Elma Schiebel and Ole Moller Jensen. De Nully Brown and Olsen were to reappear on the Nordic leukaemia study in 1992. The intention and purpose of the paper was clear from the final sentence in the abstract: Our descriptive data suggest that environmental exposures do not play any significant role in the aetiology of the majority of childhood cancers. Can you believe these people? What causes the childhood cancer then? Bad luck? Witchcraft? The only real data shown that could be used to underpin this inane conclusion was a three decade logarithmic trend graph of childhood cancer and leukaemia in the ages 0-14 from 1943 to 1984. Because they included neuroblastoma (rate 0.08 per 100,000) with leukaemia (rate 13 per 100,000) they were able to squeeze the leukaemia trend into the vanishing region of the log graph. This is a common ploy when embarrassing trends are displayed. Logarithmic graphs squeeze out any changes and should never be used for trend examination. Logarithmic display always should set off alarm bells, and three decade log makes it virtually impossible to see an effect, particularly in the way it was plotted in this paper. Nevertheless, my photo-enlarging and careful measurements showed that there was a 20% increase in leukaemia between 1953-57 and 1968-72 for the 014 age group. Since this increase occurred after the fallout, it is hardly fair to say that they had shown no effect of the environment. Their result showed little resemblance to the Hakkulinen/Hansen series or to the lost data, which I painstakingly typed into a computer program and began to unravel. Interestingly, the de Nully Brown paper did not reference Hakkulinen et al, nor Hansen et al. The airbrush was out. 13.4 Johannes Clemmesen and Otto Carlsen In this search for the correct data, what no one had noticed was that there was another player in this area, Johannes Clemmesen, the original founder of the Danish Cancer registry. Clemmesen was looking for the relationship between cancer and the environmental cause. Starting in 1943, he painstakingly collected cancer data and recodred the incidence rates in Denmark, together with essays and studies of the aetiology of cancer and leukaemia in five fat
red volumes published between 1965 and1977 covering the period 1943 to 1972. These volumes Statistical Studies in Malignant Neoplasms are an amazing source of information about cancer trends and cover the longest period of any cancer registry covering a large single genetic population. In order to examine trends, you have to have all of the five volumes. The Clemmesen series had been cited by Hansen et al. but none of the other papers had mentioned it. I began to obtain them from the British Library and to copy out the leukemia data. Here at last was something I could trust. Apart from this all there were, were tables of numbers from people who were not high on my list of believable witnesses and which didnâ€™t generally agree with each other. After looking at and plotting the Clemmesen data on childhood leukaemia in the first four volumes I became stuck. The Fifth volume, needed to complete the study was missing from the Brirtish Library. I tried elsewhere, various Universities, the internet, my daughter who is at the London School of Hygiene and Tropical Medicine. No luck. It had vanished into this air. Why? It was actually possible that this was a piece of evidence available for anyone interested in checking out the effects of fallout in cancer, not just in children but also in adults. Maybe someone had removed it from the shelves of all these libraries. I had to have it. I telephoned my friend Otto Carlsen in Denmark. Could he get it from the National Library in Copenhagen? Was it there? I had met Otto Carlsen when his political party Folk-bevaegelsen mod EU invited me to Copenhagen in 2000 to speak about the threat from the Euratom 'Basic Safety Standards Directive' which permitted the recycling of nuclear waste. He lives in Aalborg, on the northern tip of Denmark 5 hours from Copenhagen by train. I emailed him my report on the issue and left it at that. A few months later, when I had largely given up, a box appeared by carrier on the doorstep. It contained all five of Clemmensens's volumes, and a letter from Otto. The letter said: Dear Chris Busby, The famous founder of the Danish Cancer Registry, now a very old man, this morning gave me two copies of his original work, one for you and one for me. I hope this allows you to decide upon the Twist that you described to me in your letter. I couldnâ€™t reach the data through normal scientific libraries since restricted admission to the data is practiced. For certain reasons springing from my original research I chose to meet the founder personally, which I do not regret. Johannes Clemmesen, though of a very high age, is a person of integrity and dedication to his life work, described in his several volume written memoirs. Therefore I of course listened carefully to his expressing worries of the actual status of the institute (which may well be grounded?) that, ' research now has to follow 10 certain health lines instead of numerous more distinctly defined disease lines, of which some are seldom and rare'. These were Mr Clemmesens words to me yesterday. Do you have an opinion on this, or does it parallel tendencies elsewhere? My question for you. I showed him my book on the investigations of the 'International Physicians against Nuclear Weapons' on which I am preparing the 2. edition, and I guess this convinced him of me and you being worthy of him handing us over his work. After this meeting I picked it up today in the cellar of the Institute, with the help of Dr Hans Storm and Secretary Inge Bilde Hansen. Now as You can see, the one five volume copy I send to you as well as the same five volumes are to be considered our personal property, and I hope they shall help the common cause of preventing the diseases from low level radioactive pollution, certainly a noble purpose of decisive importance, and certainly a longsighted work in which You for years have been the forerunner. Many things gave happened since my organisation, Folkebevaegelsen mod Unionen invited You to our country and You spoke in our Parliament, and I am sorry to say, that especially the Nordic countries have a lot more to be grateful for in regard of your research than has been outspoken. We ourselves had excellent scientists whose work had for years to wait to meet acknowledgement. Niels Bohr is the best known. Me and my organisation wish You progress in Your work to test and refine your theory, Yours, Otto Carlsen Aalborg Oct 14th 2003
Well there it was, the whole dataset, one that I could believe since it was constructed printed and published before the arguments about Sellafield leukemias after the Black committee in 1984 and well before the trial in 1993. What did it show? Was there an increase in childhood leukemia in children or in infants in the Danish series? The answer is yes to both. And indeed, the increase is also there in the basic data that was found by Sarah Darby and arrived via the BMJ. In Fig 13.4.1 I show this data from the Clemmesen series for the 0-4year olds and plot it on top of the graph for the 0-4 years olds given by Darby et al in the Nordic leukaemia study paper in the BMJ. There is also an increase in infant leukaemia, a very significant point since we saw infant leukaemia after Chernobyl in England and Wales and the Nordic study was advanced in CERRIE as a reason why we should not believe the several studies that reported the increase in different countries. The big fat peak that is there in Denmark has been absorbed by mathematical magic into the Nordic trend shown by Darby at al. I still can't figure out how they did it as Sarah Darby hasnâ€™t given me the SAS program that they used but there you are. 13 5 Was there an increase in childhood and infant leukemia after the weapons fallout? Childhood leukaemia 0-4 is now believed to follow foetal exposure or exposure to the sperm. This is why it peaks in the age group 0-4. The increase of about 35% in childhood leukaemia follows a difference in absorbed dose to the foetus of about 80 ď Sv so if this was then we can compare the increase with the 40% increase found by Alice Stewart after 10mSv obstetric X-rays, which is now believed to be the best evidence of childhood cancer risk following exposure. The difference is about 125 times. That is to say that the Danish data suggest that 125 times more children developed leukaemia than would even be predicted by Alice Stewart's risk factors, never mind those of the ICRP which are much lower.
Fig 13.4.1 Rates of childhood leukaemia 0-4 in Denmark 1943-77 from Clemmesen publications (line with large circles) together with Darby et al Nordic leukaemia study (1992) published trend (full line). Also shown is the acute leukaemia trend of Hakkulinen et al 1986 (line with small circles) But I am not really that interested in the calculation, because this involves more arguments about which dose, which risk factor, which age group, endless roundabouts of argument and counter argument which have been brought to CERRIE by the advocates of the nuclear industry. All very tedious. All I wanted to report here was merely the way in which the
Nordic leukaemia study was constructed to show that there was no increase in childhood leukaemia following the weapons fallout when the data showed there actually was. In fact, there was a very good study in England (Bentham and Haynes 1994) that showed that there was at least a 25% increase in childhood leukemias between wet and dry areas of the country, and this result is in fairly good agreement with what the Clemmensen data and the Darby raw data from Denmark show. What of the other Nordic countries; what do they show? Well I now have the data from the other Nordic countries also, but unfortunately they all begin their series at about the point you would expect there to be an effect. If you examine each country separately, then there are various peaks which occur at slightly different times and the overall average is smudged out. The best we can do is look and see what the other Nordic countries show after 1959, the earliest year for data for the whole bloc. When this is done (as I showed in Wings of Death) there is an increase there of about 25%. Finally, when all this was revealed in the CERRIE committee, after all the work I did and the scurrying about by Otto, and the pressure from the BMJ, what did Dr Darby and Professor Doll say? They said (and this is the honest truth, knock me down with a feather, you can ask them), we knew all along that there was a peak in the childhood leukaemia incidence in Denmark. That is why we did the study: so as to see if it was there in the other Nordic countries also. And it wasn’t, so it must have been a statistical blip. Well, there you are. End of story. So why did you not say this in the paper, and cite Clemmensen, Hakkulinen et al., Hansen et al? Richard Bramhall jokingly composed a little Haiku which he called Hakku Haiku. It goes: Hakkulinen sees Fallout brings disease; later Doll cooks the data There are, of course, many problems with looking for an increase at the time of the fallout, 1959 to 1963. Lots of changes were occurring during the course of the time series. The fallout didn’t arrive in one peak like it did with the Chernobyl epidemiology. There were large changes in birth rate in the early 1960s following the sexual revolution. There was a very cold winters in the 1960s (due to the fallout in the stratosphere blocking out the sun, and the radiation affecting the ozone layer). There were many sources of radiation exposure in the late 1940s and early 1950s that would have increased the rates of childhood cancer in the early control period. e.g. obstetric X-rays, Chest X-rays, Radium dials on watches and wartime equipment like compasses. These radium dials were horribly radioactive and were clipped to the belt, giving a large dose to the testes. After the war, they were mostly sold in Army Surplus stores or just thrown away. I expect they gave much delight to children who acquired them and watched them (as I did) glow in the dark. The Army compass I still have (and use as a calibration source for our spectrometer) gives 50µSv per hour external dose. The large bubble in the card damping fluid is pure Radon. In one day, clipped to the belt, you get a whole years equivalent of natural background. Perhaps this is the cause of the sharp increase in childhood leukemia in the Shetland Islands during the influx of servicemen during the war, rather than the population mixing which Prof Leo Kinlen has based it on. The increase in child leukaemia has been inexorable since the beginning of the century. This is a strong argument put forward by Sir Richard Doll is that the radioactive releases from fallout and nuclear sites cannot be the cause of the increase in childhood leukaemia. But I complete this section with a graph I constructed following a suggestion by Richard Bramhall. In Fig 13.5.1 I plot world radium production (which is the same trend as uranium production) and childhood leukaemia mortality (which is the same as incidence up to 1960). Notice anything? We know now what is the basis of childhood leukaemia. It involves genetic damage in utero or at conception. The genetic damage causes chromosome double strand breaks. These rearrange and translocate to form new hybrid chimeric chromosomes which have sequences that code for new chimeric proteins. The translocations have been characterised and are found in child leukaemia cells (Weiss reference). It is radiation that causes double strand breaks more than any other agent, particularly alpha radiation, or
multiple tracks of beta or Auger radiation. The story is complete. It is only political power that forces it off the explanation list.